RFR: 8256265: G1: Improve parallelism in regions that failed evacuation [v4]

Thomas Schatzl tschatzl at openjdk.org
Wed Sep 14 09:24:54 UTC 2022


On Tue, 13 Sep 2022 13:43:13 GMT, Albert Mingkun Yang <ayang at openjdk.org> wrote:

> Instead of calling `note_end_of_clearing` in `G1ConcurrentMark::clear_bitmap_for_region`, I wonder if moving it (`note_end_of_clearing`) to a higher-level caller (`G1ClearBitmapClosure`) is a bit nicer. This way, `G1ConcurrentMark::clear_bitmap_for_region` does exactly what its name suggests, nothing more&less.

Here's a branch that explicitly sets TAMS: https://github.com/openjdk/jdk/compare/master...tschatzl:jdk:submit/8256265-parallel-evac-failure-alt?expand=1

The last commit (https://github.com/openjdk/jdk/commit/b77bb2f46d144345bdcb7b70156eba6491a3d00b) implements the suggested change.
This omits that call in the second invocation of `G1CollectedHeap::clear_bitmap_for_region()` because TAMS must be `bottom()` at that point.

I do not think this difference is the heart of this review and should not be preventing progress :) Let's postpone the discussion whether TAMS should be part of `G1ConcurrentMark` or not to another day. So I would be happy to use that going forward for this review.

-------------

PR: https://git.openjdk.org/jdk/pull/9980


More information about the hotspot-dev mailing list